The Gain in a GainClone?

[Mmne]

Im sitting here waiting for my samples of lm4766/4780 to arrive searching the net for info about design ideas and a question arrives in my mind. Almost all the designs are "typical application" design of the datasheet and in this circuit has a gain of 21. I wonder if all that gain is nessesary? I have seen discussions where people claim that too much gain in a stage is bad for the musical quality. The standard approach to OP-amp design is also: do not use a voltage divider for volume adjustment, use variable gain instead. Have I missed something here?

[cpemma]

You can set your gain according to your needs (power rails and source level); 21V/V works out nicely for +/-23V rails (eg, from a 2 x 18V AC transformer) and a +/-1V peak source...

And the problem with variable gain is keeping the feedback path short and clean, plus I don't think those chips are stable at low gain.

[Mmne]

It is proably as you say, its not stable at to little gain, but I cant find anything about minimum gain for stable operation in the datasheet for lm4766. I stick for the simple solution and go for the exampledesign, it is my first one and the tweaking can wait to later versions. I'm going to use a 225VA,2x18V torroid and 2x4700uF caps on each rail. What do you think about that? Overkill? I am going to use it for my nearfield monitors, fullrange 4 ohms.

[I_Forgot]

Originally posted by Mmne
I have seen discussions where people claim that too much gain in a stage is bad for the musical quality. The standard approach to OP-amp design is also: do not use a voltage divider for volume adjustment, use variable gain instead. Have I missed something here?

I don't know who that sort of thing is standard with, but if they have audio in mind, it isn't a very good thing to do. An op-amp has a fixed gain-bandwidth product. If you reduce the gain, you increase the bandwidth, and vice-versa. In an audio system you generally want to limit the bandwidth to the audio range. If you operate at low gain and very wide bandwidth you run into problems with interference from radio stations, switch mode power supplies, etc.

The conventional method of connecting a volume control between the input signal and ground with the op-amp input connected to the wiper is a good way to do it.

I_F

[peranders]

Originally posted by Mmne
OP-amp design is also: do not use a voltage divider for volume adjustment, use variable gain instead. Have I missed something here?

It seems like a good idea but look around: why can't you see this solution more often?

Fxed gain will give you more predictable results and is more reliable.

[Mmne]

I definetly see your points, interesting. The way to go I have learned is to use as little gain as possible beacuse having more gain than necessary will increase noise and for the bandwith reduction I use HP/LP filtering. But my background is not HIFI. I entered the world of electronics tweaking guitaramps and building stompboxes, so this is a new chapter for me.

[peranders]

If you have gain of 20-30 most parameters will be good enough, > 100 kHz bandwidth, dist < 0.005%, noise very low...

You are right about to have not more gain than needed but do you have more than 1 V out to drive the amp?

[Mmne]

>You are right about to have not more gain than needed but do you have more than 1 V out to drive the amp?

No I don't think so, this first one is going to be connected to the outputs of my soundcard, I'm not sure but I think it is nominal line voltages. But if I'm pleased with the result I'm going to build another to use with my stereo and then I want a preamp of some sort and then the extra gain in the power stage may be unnecessary.

[Zero One]

I have built 10 gainclone modules up to now and I am working on a very small LM1875 chip based clone, housed in a computer CD drive case. One of the things I was considering was lowering the gain say to 18. The circuit is very compact, using SMDs soldered directly to pins etc where possible with the common ground just next to the nut and bolt going through the chip.

Each module is encased in a small alloy cube within the case with power and inputs and outputs exiting through the alloy plate and joining virtually directly to the pins on the chips, and inputs and outputs and power are all isolated from one another, one wall acts as the heatsink. I am assuming this will limit possible feedback issues, so far I have built one experimental module to test things although I did build a similar and successful gainclone headphone amp (canclone) this way that ran on a 5v smps supply (sounds silly I know but it works really well).

This clone will run on either a 12v smps or 16V snubberized supply and be fed through a really good quality preamp that I have almost finished.

Anyhow that's the background as I said I was interested in the gain issue, I wonder if a lower gain will improve resolution, distortion, noise floor etc, I am intersted to know what others have experienced here. Is the gain for example deliberately set high in order to avoid feedback problems/stabilty or is it because the amps actually perform better this way. Nat Semis info sheets really don't give much away on the issue. Frankly I'm no tech head where electronics is concerned, I only built my first amp back in late January, I just tinker a lot so any input is really helpful to me.

I've tried many things with my clones and some really do make substantial difference, but this one doesn't seem to get much coverage. Overall I am interested in getting absolute maximum detail from the amps and a really good top-end, LM1875s and 12V smps power goes a long way towards this (pretty tube like to my ears) from my experiments but is there still more lurking in there somewhere.

[cpemma]

Originally posted by Zero One
I have built 10 gainclone modules up to now and I am working on a very small LM1875 chip based clone, housed in a computer CD drive case. One of the things I was considering was lowering the gain say to 18...

the datasheet
The amplifier is internally compensated and stable for gains of 10 or greater.

And with +/-12V rails there's no point in using the 21V/V gain if your preamp is up to scratch. See if the chaps at NS are telling it as it is.